Fake Review Detection in E-Commerce Using Machine Learning and NLP: A Comparative Study

Abstract: The proliferation of fake reviews in e-commerce platforms has become a critical challenge affecting consumer trust and purchasing decisions. This paper presents a comparative study of machine learning approaches for automatic fake review detection using Natural Language Processing (NLP) techniques. We evaluate three classification algorithms — Logistic Regression, Naive Bayes, and Support Vector Machine (SVM) — on a dataset of 8,087 reviews using TF- IDF feature extraction. Experimental results demonstrate that SVM achieves the highest accuracy of 86.60% with precision and recall of 0.87, outperforming Logistic Regression (86.01%) and Naive Bayes (84.37%). The proposed system effectively distinguishes between genuine (CG) and fake (OR) reviews, providing a reliable foundation for trustworthy product recommendation systems.

Keywords: Fake review detection, Natural Language Processing, TF-IDF, Support Vector Machine, E-commerce, Opinion spam.